On 2008-10-13 18:22+0200 Robert Pollak wrote:
> Hi Alan,
>
>> I assume you are referring to the kind of plot that is discussed in
>> http://en.wikipedia.org/wiki/Box_plot.
>
> Yes, that's what I mean.
>
>> We currently have no higher-level API to support that. Of course,
>> you could make your own function to do that using lower-level PLplot
>> commands.
>
> I am now drawing such a box plot. I also want to mark outliers with
> circles. Which aspect ratio do I need to draw undistorted circles?
> (In other words: How to map world to device coordinates?)
>
> As I don't want to use "o" letters [...]
<aside>
We actually have access to circular symbols (better than the "o" character)
that are not subject to aspect ratio problems so I tried some experiments
with those using the following python test script.
********
#!/usr/bin/env python
# Append to effective python path so that can find plplot modules.
from plplot_python_start import *
import sys
from plplot import *
from numpy import *
# Parse and process command line arguments
plparseopts(sys.argv, PL_PARSE_FULL)
# Initialize plplot
plinit()
# Like yellow lines better.
plcol0(2)
pladv(0)
plvpor(0.1, 0.9, 0.1, 0.9)
plwind(0., 1., 0., 1.)
# Just to show edges of viewport
plbox("bc", 0., 0, "bc", 0., 0)
x = [0.5]
y = [0.5]
ifplsym = False
ifunicode = True
for size in 1.0+arange(10):
if ifplsym:
plssym(0., size)
plsym(x, y, 907)
else:
if ifunicode:
plschr(0., 2.*size)
plptex(0.5, 0.5, 1., 0., 0.5, "#[0x25ef]")
else:
plschr(0., size)
plptex(0.5, 0.5, 1., 0., 0.5, "#(907)")
# Terminate plplot
plend()
********
To see results from that script you have to run it in the examples/python
directory within the installed directory tree (not the source or build
tree).
The Hershey symbol version (using plsym [ifplsym == True]) actually gives a
pretty good-looking circle, but you can probably make something that looks
smoother if you use more points to represent it. Thus, I don't think the
plsym version is quite suitable for your needs although it comes close so
you should at least try it. The plptex version without unicode gives the
same result as the plsym variant for drivers that use the Hershey fonts, but
for drivers that use Unicode fonts (ifunicode == True), plptex has an
obvious bug (repeated strings in the same position are not plotted in the
same position) and the magnification done with plschr also magnifies the
width of the line in a way that might make it unsuitable for your needs.
In sum, plsym might give you what you need, but you will have to check, while
plptex has a bug currently, and furthermore, even when that is fixed the
magnification of the line width may be a concern. Anyhow, try some of
these variants and see whether there are any of them that you like (assuming
the positional bug gets fixed).
</aside>
> I tried the following, but it gives
> me ellipses, e.g. when called in x12c.c:
>
> plcircle(PLFLT x, PLFLT y)
> {
> PLFLT vpxmin, vpxmax, vpymin, vpymax;
> PLFLT ratio, rx, ry;
> int i;
> static PLFLT cx[31], cy[31];
>
> /* get axis ratio */
> plgvpw(&vpxmin, &vpxmax, &vpymin, &vpymax);
> ratio = (vpxmax -vpxmin) / (vpymax -vpymin);
>
> ry = 1.;
> rx = ratio * ry;
> for (i = 0; i <= 30; ++i) {
> cx[i] = x + cos(i / 30. * 2. * M_PI) * rx;
> cy[i] = y + sin(i / 30. * 2. * M_PI) * ry;
> }
>
> pllsty(1);
> plline(31, cx, cy);
> }
>
> (I also don't understand the meaning of plgvpd(). What exactly are
> "normalized device coordinates", e.g. when I use the wxwidgets device?)
I suggest you review
http://plplot.sourceforge.net/docbook-manual/plplot-html-5.9.0/viewport_window.html
to get definitions of what you need to know and lots of other useful links
to documentation of individual commands. However, it's pretty clear to me
this documentation could use a rewrite. So if after digging through this
documentation and comparing with what actually happens in the code, you have
some improved wording to suggest, please let me know. A patch on
doc/docbook/source/*.xml is the best form to communicate suggested changes
to our documentation, but ascii is OK as well.
To answer your general question about aspect ratio, plgvpw is not suitable
for your needs because it returns world coordinates and not device
coordinates. From the definition of "normalized device coordinates" in the
above URL, it appears plgvpd is also not suitable for your needs.
What you appear to need is something that returns real device units (whether
they are in pixels, points, or mm doesn't matter since it is only the ratio
the determines the aspect ratio). I therefore suggest you try plgspa or
plgpage. However, I haven't tried those, and I think it is probably best
that you survey all the modern device driver codes to be sure everything you
need will be defined for whichever command (plgspa or plgpage) that you
decide to use.
Alan
__________________________
Alan W. Irwin
Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).
Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the libLASi project (unifont.org/lasi); the Loads of
Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________
Linux-powered Science
__________________________
-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
Plplot-general mailing list
Plplot-general@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/plplot-general
